quickstep-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From zuyu <...@git.apache.org>
Subject [GitHub] incubator-quickstep pull request #318: Optimizing Hash repartition on a raw ...
Date Wed, 18 Oct 2017 22:57:10 GMT
GitHub user zuyu opened a pull request:

    https://github.com/apache/incubator-quickstep/pull/318

    Optimizing Hash repartition on a raw pointer, instead of using TypedValue.

    This PR optimizes hash repartition over a raw pointer, instead of `TypedValue`.
    
    Experimental results show that repartition a `lineorder` from `SSB-100` on would reduce
to `750 ms` from `1000 ms`.
    
    Note that this PR allows supports the single partition attribute. For multi-attribute
repartition, there are two basic approaches.
     1. like the single partition optimization, combine per-column hash, and then set the
bit vector for the result partition.
     1. per-tuple evaluation, w/o any extra buffer used in the above approach.
    
    The first approach on the single partition attribute actually takes `1200 ms`, while the
latter takes `850 ms`.

You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/zuyu/incubator-quickstep hash-p-void-single

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/incubator-quickstep/pull/318.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #318
    
----

----


---

Mime
View raw message